TD posted on Wednesday, November 28, 2007 - 7:25 am
I am trying to conduct two CFAs with the same variables - 21 indicators overall, many of which are ordinal. The first is a 4 factor CFA, the second is a higher order CFA using the four factors from the first CFA. However, I have problems with each when I treat the variables as ordinal/categorical and use WLSMV as the estimator. In the first CFA, the program does not appear to converge even after an hour of calculation time. With the higher order CFA, the estimation appears to terminate but I get no results - I assume estimation did not terminate normally. The programs seems to run well when I treat the all of variables as continuous and use MLR as the esitmator, but I know this is not a prudent solution. Can you offer any explanations and/or suggestions?